Simply connect the negative input of the operational amplifier (which must be able to go
both positive and negative) to the junction of RS1 and RS2, and connect the voltage-
setting resistor of the LM117 to the output of the op amp, instead of to ground.
The dual-tracking stability which will be observed will be about twice the input offset
variation (over time and temperature) of the operational amplifier used. Since dual-
tracking to much better than 1 mV amounts to overkill, any op amp which has half-way
decent input offset stability should work. An OP-07 is better than required; even a 741
would be o.k. However, please refer to the precautions against oscillation which are
mentioned on the next page.
The op amp stabilization system is only intended to handle relatively slow regulator
excursions or drifts, so a high-speed amplifier is not needed. In fact, if the op amp were
too fast it might interact adversely with the normal desensitizer net, which must still be
used to handle the effects of fast transient regulator excursions (such as might otherwise
result from, e.g., a 20-
sec high-power pulse).
We have, in fact, observed that the optimum setting of RS3 in the desensitizer net is not
quite the same when set for cancellation of the effects of a DC voltage shift as it is when
set for cancellation of the effects of a fast voltage transient. As a result, the shape of the
back end of a long high-power pulse may sometimes be further improved by adjusting
RS3 to eliminate any residual effects of such transients on the baseline, as illustrated
below in Fig. C2.
Fig. C2
Tuning the desensitizer to best deal with transient regulator effects will usually enable the
use of smaller decoupling caps. This can be considered a secondary benefit of the op
amp stabilization approach, and might be particularly useful for hybrids, since the op amp
die are very small.
-C2-
 |
 |
 |
 |
 |